Cyclic nucleotide-gated (CNG) channels play essential roles in the transduction of visual and olfactory information (Stryer, 1986;Zufall et al., 1994). They sense variations in the intracellular concentration of cyclic nucleotides that occur in response to visual or olfactory stimuli. In many ways, CNG channels are similar to voltage-activated potassium (Kv) channels. They coassemble as tetramers of homologous subunits (Weitz et al., 2002;Zheng et al., 2002;Zheng and Zagotta, 2004;Zhong et al., 2002), each containing six transmembrane segments (TM), a positively charged TM4 and a reentry P-region between TM5 and TM6, suggesting that CNG channels belong to the same superfamily of voltage-activated cation channels (Jan and Jan, 1990). The main difference is that CNG channels are not gated by changes in membrane voltage. Instead, they open and close the pore in response to changes in the intracellular concentrations of cGMP or cAMP, a property conferred by the presence of a cyclic nucleotide binding domain at the C-terminus of each subunit (Kaupp and Seifert, 2002;Zagotta and Siegelbaum, 1996).Our understanding of how CNG channels open and close their pore in response to cyclic nucleotide binding is much less refined than our understanding of how Kv channels gate in response to voltage. A large body of evidence, using a variety of approaches, has established that Kv channels open and close their permeation pathway at the intracellular end of the pore (Armstrong, 1971;Armstrong and Hille, 1972;del Camino and Yellen, 2001;Doyle et al., 1998;Holmgren et al., 1997;Jiang et al., 2002;Liu et al., 1997;Perozo et al., 1999). Attempts to extend those ideas to CNG channels have encountered some resistance. For example, studies using molecules that block the permeation pathway of CNG channels, such as tetracaine (Fodor et al., 1997a;Fodor et al., 1997b) or quaternary ammonium ions (Contreras and Holmgren, 2006) have shown that blockade is not state dependent, as if these molecules can access the pore in both open and closed channels;which is in stark contrast with the blockade properties observed in Kv channels (Armstrong, 1966, 1969;Armstrong, 1971;Armstrong and Hille, 1972;Choi et al., 1993;Holmgren et al., 1997). In addition, experiments examining the state-dependence of cysteine modification by intracellular application of methanethiosulfonate (MTS) reagents have failed to show dramatic differences between open and closed states in the inner vestibule region (Contreras and Holmgren, 2006;Flynn and Zagotta, 2001;Sun et al., 1996), results that are inconsistent with an intracellular gate in TM6, as shown in Kv channels (del Camino and Yellen, 2001;Liu et al., 1997). Several studies indicate that the pore region of CNG channels play a role in gating. For example, accessibility of cysteine reagents applied from the intracellular (Becchetti et al., 1999;Becchetti and Roncaglia, 2000) and the extracellular (Becchetti et al., 1999;Becchetti and Roncaglia, 2000;Liu and Siegelbaum, 2000) side of the channel to cysteines substituted along the entire P-region have shown that modification of some residues in this region perturb normal gating by cGMP. Interesting, by measuring modification rates in different states of the channel, Liu and Siegelbaum (2000) were able to show that the pore helix undergoes conformational changes associated with gating. The pore helix, however, does not line the permeation pathway (Doyle et al., 1998;Long et al., 2005;Shi et al., 2006). Hence, the location of the gate remains to be determined. Recently, the crystal structure of a bacterial non-selective cation channel has been solved (Shi et al., 2006) and the structure of the P-region of this channel has been proposed to be equivalent to that of CNG channels. Using this structure as a model, we have introduced cysteines along the selectivity filter as targets for chemical modification with small cysteine reagents like Cd2+ and Ag+ applied from the intracellular side of the channel. We observe that along the selectivity filter there is a state dependent pattern consistent with the idea that CNG channels gate at this region, where some positions are readily accessible in the open and closed states, while others are accessible in the open state with much slower modification rates in the closed state, similar to observations along the intracellular end of TM6 in Kv channels (del Camino and Yellen, 2001;Liu et al., 1997).